After a brain cancer patient starts treatment, the critical question is: how well is the treatment working? A method similar to forecasting storms with computer models has been developed to predict an individual patient’s brain tumor growth. This growth forecast will enable physicians to rapidly identify how well the tumor is responding to a particular therapy. The approach allows a quick pivot to a new therapy in a critical time window if the current therapy is not effective.
Brain cancer patients greatly need an approach to find optimal personalized treatments. Brain tumors vary in their growth rate, shape and density but existing methods for measuring a treatment’s impact ignore this variation. The methods (and thus physicians) cannot distinguish between a patient with a fast-growing tumor that responds well to treatment and a patient with a slow-growing tumor that responds poorly. By using a personalized, patient-specific approach that accounts for tumor features such as 3-dimensional shape, density and growth rate, this new method can make this distinction.
“There is this muddy zone right after the first round of treatments when it’s hard for the clinician to know whether to change therapy because she doesn’t have the metrics that correlate to outcome,” said Kristin Swanson, MD, of Northwestern University Feinberg School of Medicine. “The doctor can’t yet gauge how much it helped.”
If the doctor determines the treatment isn’t effective, she can try a different type of treatment or help the patient enroll in a clinical trial with a new drug being tested. The information also is helpful to the patient. “The patient wants to know the therapy is doing something for them,” Swanson said. “On the flip side, if the therapy isn’t helping, then it may not be worth the side effects he is enduring.”
This study involved 33 patients with glioblastoma, which is the most common and aggressive form of brain cancer. The study was published in PLoS ONE (2013; doi:10.1371/journal.pone.0051951).
Like weather models for the approach of a hurricane, this brain tumor model forecasts where the tumor is going, how much and where it will grow, explained Swanson. That information allows the researchers to know how much the treatment deflected the growth and to directly relate that to its impact on patient survival.
To measure a treatment’s effectiveness, the scientists performing the study created a unique computer model of each patient’s tumor and predicted how it would grow in the absence of treatment. The prediction model was based on MRI scans taken on the day of diagnosis and on the day of surgery. The difference between these two scans enabled researchers to estimate how fast the tumor was growing along with the density of tumor cells throughout the brain. Researchers then scored the effectiveness of the patient’s treatment by comparing the size of the patient’s tumor after treatment to the model-predicted size if untreated.